Deep Learning for Electrical Impedance Tomography

Collaborations

2019-Present. Univ Clemson and University of North Carolina (UNC) at Charlotte (Prof. Taufiquar Khan)

We are interested in solving the exponentially ill-posed electrical impedance tomography (EIT) inverse problem using a newly developed deep neural network algorithm. Electrical impedance tomography is an effective, non-ionizing, and inexpensive imaging modality for imaging the electric conductivity within an object from a set of boundary data. In practice, reconstructed images from traditional deterministic methods provide poor resolution. Our deep neural net algorithm was developed using convolutional neural networks for the direct problem and a deep network for the inverse problem to provide higher resolution images

Publications

. Electrical Impedance Tomography using Sparsity Regularization and Machine Learning Approach.. In SIAM Conference on Imaging Science (IS20), 2020.

PDF Project bibtex